tools/power: turbostat: fix large c1% issue
Under some conditions, c1% was displayed as very large number,
much higher than 100%.
c1% is not measured, it is derived as "that, which is left over"
from other counters. However, the other counters are not collected
atomically, and so it is possible for c1% to be calaculagted as
a small negative number -- displayed as very large positive.
There was a check for mperf vs tsc for this already,
but it needed to also include the other counters
that are used to calculate c1.
Signed-off-by: Len Brown <len.brown@intel.com>
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index b815a12..861d771 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -444,6 +444,9 @@
old->c7 = new->c7 - old->c7;
}
+/*
+ * old = new - old
+ */
void
delta_thread(struct thread_data *new, struct thread_data *old,
struct core_data *core_delta)
@@ -482,19 +485,20 @@
/*
- * As mperf and tsc collection are not atomic,
- * it is possible for mperf's non-halted cycles
+ * As counter collection is not atomic,
+ * it is possible for mperf's non-halted cycles + idle states
* to exceed TSC's all cycles: show c1 = 0% in that case.
*/
- if (old->mperf > old->tsc)
+ if ((old->mperf + core_delta->c3 + core_delta->c6 + core_delta->c7) > old->tsc)
old->c1 = 0;
else {
/* normal case, derive c1 */
old->c1 = old->tsc - old->mperf - core_delta->c3
- core_delta->c6 - core_delta->c7;
}
+
if (old->mperf == 0) {
- if (verbose) fprintf(stderr, "cpu%d MPERF 0!\n", old->cpu_id);
+ if (verbose > 1) fprintf(stderr, "cpu%d MPERF 0!\n", old->cpu_id);
old->mperf = 1; /* divide by 0 protection */
}